1. 1 SNAP/JDEM Physics and Detector R&D at Caltech HEP Alan Weinstein and Justin Albert in collaboration with Caltech Astronomy (R.Ellis, J.Rhodes, others), Caltech Optical Observatory detector group (R.Smith, M.Bonati, others), the SNAP Collaboration, Caltech SURF students Jeff Naecker, Lauren Porter, Aliza Malz Caltech HEP DOE Review Monday, August 7 th 2006

2. 2 SNAP / JDEM and Caltech HEP SNAP Collaboration is designing and developing a wide-field space telescope to perform precision measurements of cosmological parameters (dark matter & dark energy), using complementary techniques of  History of the Hubble expansion using SNe Ia as quasi-standard candles  Evolution of dark matter clustering using Weak Gravitational Lensing DE science, and SNAP in particular, has been emphasized by DOE HEPAP P5, the DETF, NASA reviews, and House and Senate FY07 Energy&Water bills. SNAP prepared a NASA ROSES proposal for JDEM in March 2006, reviewed as “Excellent” Weinstein joined SNAP Collaboration (Associate Member) in 2005  NIR detector R&D, with Caltech Optical Observatory (COO), JPL, others  Weak gravitational lensing, with COO (DOE-funded, R. Ellis)  Development of SNAP simulation package, with SNAP collaborators No DOE funding yet  propose support for AJW (1/3 of summer, redirected), one postdoc and one grad student  proposal submitted December 2005

3. 3 SNAP at Caltech SNAP project provides substantial funding for research at Caltech  WGL (Richard Ellis and Jason Rhodes)  NIR detector development (Roger Smith) Since joining SNAP I have forged collaborations with both groups The NIR group at Caltech is quite productive, but they are all technical personnel (relatively expensive!). I’ve been supporting undergraduate researchers on department money, but we need more involvement from scientists – postdocs and students, to leverage the technical expertise in Roger’s group. SNAP project management is supportive and enthusiastic about deeper Caltech involvement. Now is the time to begin to build up a strong science-based group, as is done by most other SNAP collaborators (Michigan, Indiana, Yale…). DOE support is key!

4. 4 Collaborators Work on Weak Gravitational Lensing, with:  Caltech HEP BaBar postdoc Justin Albert (  U Victoria)  Caltech Optical Astronomy (R. Ellis, J. Rhodes, R. Massey, W. High)  Closely associated with Alexandre Refregier (CEA Saclay)  Caltech SURFs Mark Rodgers (Cambridge U), Lauren Porter, Aliza Malz Work on SNAP NIR detector development / testing, with:  Caltech Optical Astronomy (R. Smith, M. Bonati, D. Guzman)  MOU between SNAP R&D project (DOE) and COO  SNAP collaborators (U Michigan, JPL, NASA GSFC)  Caltech SURF Ben Olsen (Caltech ), Jeff Naecker (Berkeley)

5. 5 SNAP/JDEM Mission Design Mission: Wide-field imaging from space ~2 meter aperture telescope to reach very distant SNe and background galaxies 0.3 deg 2 mosaic camera, 600 Mpix, 9 fixed filters for precision photometry on large numbers of SNe 0.4 – 1.7 um spectrograph for detailed study of each SN L2 halo orbit limited only by zodiacal light Deep field scan (SNe): 7.5 deg 2 scan in each of North & South ecliptic poles, in each of 9 filters, revisited every ~4 days Wide field scan (WGL): ~300 deg 2 per year

6. 6 Focal Plane Layout with Fixed Filters Optical CCDs: 441 Mpix; 0.10 arcsec/pixel NIR hybrid detectors: 151 Mpix, 0.17 arcsec/pixel FoV: ~0.3 square degrees to match CCD FoV, observe SN in every color Wavelength coverage 0.9 – 1.7  m with nine optimized filters, to observe V band out to z = 1.7 Calibration for precision photometry, photo-z 4-fold rotational symmetry to optimize scan strategy with minimal spacecraft re-orientation

7. 7 Weak Gravitational Lensing  Weak distortion of background galaxy shapes due to foreground DM clusters  Direct measure of the distribution of mass in the universe, as opposed to the distribution of light, as in other methods (eg. Galaxy surveys)  Measure lensing as function of angular size on the sky, and redshift: evolution of mass clustering (changing expansion rate due to dark energy) Distortion Matrix: Shear map:

8. 8 Measuring Weak Lensing with SNAP z S < 1.0 z S > 1.0 SNAP wide (300 deg 2 ) Rhodes et al. 2003, Massey et al. 2003, Refregier et al. 2003 Dark Energy equation of state: w=p/  (w=-1 for  ) modifies: angular-diameter distance growth rate of structure power spectrum on large scales (Ma, Caldwell, Bode & Wang 1999) w can be measured from the weak grav lensing power spectrum  SNAP will measure the evolution of the lensing power spectrum and is sensitive to the non-linear evolution of structures

9. 9 Weak Lensing with Current Data  Correlation with distribution of light provides info on dark matter properties  GOODS analysis (Albert, Rodgers, Weinstein et al 2005-2006)  In preparation for SNAP, we are contributing to the development of an analysis pipeline for extracting cosmological parameters from WGL, with tight control of systematic errors.  Development and testing of algorithms on currently available data yield measurements of mass distributions and power spectra, and provide a baseline for predictions for SNAP.  We have measured the gravitational shear and the mass distribution for the GOODS and COSMOS HST surveys (0.1 and 1 sq. deg. respectively; too small for cosmology!).

10. 10 Weak Lensing with Current Data  The GOODS survey is an HST survey in 4 bands covering 0.1 sq arcmin. of the north and south ecliptic poles, taken in 2003-2005.  Survey taken in 5 “epochs” combined to increase survey depth and remove noise.  Precise corrections that are uniquely important for WGL (CTE, photometric…) in progress. Results from GOODS (and COSMOS, …) should follow shortly! HST GOODS north ecliptic tiles GOODS south ecliptic tiles Typical data image

11. 11 errors (measurement  cosmology) TRUE cosmology TRUE WGL power spectra (z bins) random WGL shear map (z bins) measured WGL shear map (z bins) measured WGL power spectra (z bins) measured cosmology theory HealPix syst errs: eg, photo-z simulations shapelets HealPix FisherMatrix (DETF) add errors stat analysis Under development (Weinstein, Porter, Malz) Rhodes et al REAL stat and syst errs

12. 12 Optical Bands NIR Bands Z = 0.8Z = 1.2Z = 1.6 Simulated SNAP observations of high redshift SNe NIR observations in SNAP science NIR observations in SNAP science NIR allows observation of SNe rest frame optical to high redshift Tracing the effect of dark energy through cosmic time requires probing to high-redshift Rest frame optical shifts into NIR after z=0.9 Large wavelength coverage provides important constraints on systematic errors (eg, dust extinction) Substantially enhances auxiliary science Visible Only Visible + 3 NIR Prior on  m

13. 13 NIR Science drivers Spatial and wavelength coverage: NIR data for all SNe to constrain systematic errors  focal plane to match CCDs which cover the visible Wavelength coverage to overlap CCDs and allow B and V restframe observations to z>1.5  sensitivity from 1.0 to  1.7  m Signal-to-noise ratio: Noise should be dominated by unavoidable zodiacal light  dark current < 0.02 e - / pixel / sec  read noise < 5 e - Signal levels be sufficient to allow precise observation of SNe near peak with adequate S/N out to z=1.7 within time constraints  quantum efficiency > 60% Gain stability and linearity sufficient for precise photometry Initial SNAP Spec NIR detectors have (somewhat unexpectedly) high QE, permitting more relaxed read noise requirements. The read noise has also been improving considerably with recent sample detectors from both manufacturers. These requirements can be met!

14. 14 NIR detector R&D Establish large format detectors with good QE out to 1.7 µm cutoff, with low read noise, operating at SNAP FPA temperature of 140 o Explore broadened technology options and vendor pool; establish competitive vendor environment  Rockwell (RSC) – MBE HgCdTe  Raytheon (RVS) – LPE HgCdTe  Sensors Unlimited/Rockwell – InGaAs Establish facilities for testing and characterizing NIR FPA and detectors (Michigan, Caltech, JPL, Goddard) Over the last 2 years the SNAP infrared program has established that detectors satisfying the science requirements (read noise, QE, gain stability and linearity) are achievable (or close) with detectors from both vendors.

15. 15 NIR 4 Mpix detector/MUX hybrids from Rockwell Scientific Detector: MBE HgCdTe 2048  2048 pixels (18  m pitch), includes reference pixels 4-edge buttable for large mosaics Indium-bump bonded to HAWAII-2RG MUX, 32 readout channels ADC / readout ASIC (SIDECAR) under development Under development for JWST (to replace Hubble in 2011…)

16. 16 Quantum Efficiency Both manufacturers can produce QE>90% Raytheon have demonstrated AR coating on other devices, and are expected to succeed on these. U Michigan

17. 17 SNAP NIR R&D at Caltech Memorandum of Understanding between SNAP R&D Project and Caltech Optical Observatories (July 2005) (Smith, Ellis, Dekany, Weinstein) NIR Detector testing: research and development on  Dark Current noise  MUX, readout controller noise  Quantum efficiency vs wavelength  gain linearity and stability  inter-pixel and inter-detector cross-talk  readout methods to minimize read noise Standardized testing for production. NIR Detector control ASIC testing and characterization. Collaboration with SNAP Calibration and Electronics working groups.

18. 18 Detector / Dewar test setup

19. 19 Linearity signal vs exp.time @ constant flux Low flux but still well above dark which was subtracted anyway High flux ADC saturation curved due to illumination variation. H2RG-038 Linear fit forces agreement here Ben Olsen, Caltech

20. 20 Dark Current, Read Noise Rockwell dark current is very low, yet for 1% of pixels, noise > 2*mode due to hot pixels Spectrograph at 120K is ok. (Rockwell) Mean Raytheon dark current is close to 0.1e-/s imager spec but there are too many hot pixels. Better devices promised soon by Raytheon. Rockwell Detector Noise > biases > mux > video Raytheon Mux Noise > detector, biases > video

21. 21 Persistence, Drift, bad pixels Persistence is the release of charge following illumination of HgCdTe arrays. Appears to be both flux and intensity dependent Mitigation through clocking, biasing… Correctable if it can be reduced to acceptable levels Zero point drift due to drift of bias voltages and other uncontrolled parameters. We are improving bias voltage stability, monitoring, immunity Use reference pixel subtraction, optimally, to minimize effect. Pixel operability: dead, hot, unstable pixels. Ex: H2RG-103 where 4% are 2x away from mode More than half are due to detector, MUX noise, Need simulations to assess impact on science.

22. 22 Compliance matrix Compliance matrix RVSRSC QE      Intrapixel response      Lateral charge diffusion (MTF)  Interpixel Capacitance  Sensitivity dependence on flux?? QE & Gain stability, linearity?? Mean dark current @ 140K  Hot pixels @ 140K  Cosmic rays (substrate removal)  Mux glow?    Persistence?  Read Noise (15s+15s fowler)  Zero point drift (biases)  Zero point drift (thermal) ? ? ?? Pixel Operability %?? Legend U  exceeds initial requirement U meets requirement ° near requirement ? not yet well characterized  unacceptable  major improvement unlikely …a simplified overview Sensitivity Dark signals Noise R. Smith, Caltech To date focused primarily on sensitivity issues Shifting to focus more on precision

23. 23 Conclusions Caltech HEP is developing a significant program in cosmology science and SNAP/JDEM R&D, working closely with Caltech astronomers and SNAP collaborators. We seek to build up our group with the addition of a postdoc and a grad student, and wish to request funding from the DOE. Weak gravitational lensing:  Development of analysis pipeline, simulation tools, systematic studies.  We are presently analyzing HST and ground-based surveys (GOODS and COSMOS), and making weak lensing measurements with present data. NIR detector development:  NIR detection is essential for SNAP science!  NIR requirements are demanding but achievable.  Development of dedicated testing, R&D facility for study and characterization of dark noise, readout noise, gain & linearity, QE, stability, readout methods, etc.  NIR detector readout ASIC testing.  Production testing procedures, facilities. SNAP will constrain dark energy properties by over an order of magnitude better than they are constrained today (and provide precision information on presently unconstrained parameters such as w′ ). Both supernovae and gravitational weak lensing can provide precision constraints from the SNAP dataset, and they provide complementary information. With a wide, deep survey such as SNAP will provide, our understanding of dark matter, dark energy, and the large-scale structure of the universe will be revolutionized.

24. 24 Dark Energy and Caltech HEP It is my intention to make significant contributions to the study of Dark Energy (via SNIa, WGL) with DOE support. It is my desire to commit to the development of a next-generation Dark Energy measurement effort which will be realized in the not-too-distant future. To this end, I joined SNAP Collaboration (associate member) in October 2004. There is much uncertainty as to the ultimate fate of SNAP, JDEM, and other promising projects (LSST, PanSTARRS-4, ADEPT, etc). SNAP is an excellent way to study Dark Energy, perhaps the best way; but it is not the only way, and it is expensive and complex; and multiple approaches are good and often complementary. I am wedded to the science, but not necessarily to the project. We can’t do the science if the project is mired in funding and politics problems! I intend to work in the context of the project that has the best chance of being carried to realization by the DOE and collaborating agencies. Guidance from the DOE would be very valuable! It is still some years before any of these next-generation experiments are realized. I plan to continue to pursue work in CMS and in LIGO for the next few years.

25. 25 SNAP/JDEM Mission Design Wide field imaging from space Mission: L2 halo orbit Deep field scan (SNe): 7.5 deg 2 scan in each of North & South ecliptic poles, in each of 9 filters, revisited every ~4 days Wide field scan (WGL): ~300 deg 2 per year

26. 26 Dark Energy equation of state: w=p/  (w=-1 for  ) modifies: angular-diameter distance growth rate of structure power spectrum on large scales (Ma, Caldwell, Bode & Wang 1999)  w can be measured from the lensing power spectrum  But, there are degeneracies between w,  M,  8 and  Dark Energy and Weak Lensing Cf. Hui 1999, Benabed & Bernardeau 2001, Huterer 2001, Hu 2000, Munshi & Wang 2002 a(t) linearnon-linear

27. 27 Proposed involvement in NIR R&D Caltech astronomy has a strong, experienced team of IR instrumenters, led by Keith Taylor and Roger Smith They are involved in many projects; SNAP involvement requires more participation. Propose to work with this team, learning the techniques and technologies; and work in close collaboration with other groups pursuing parallel goals (Michigan, JPL, GSFC). SNAP effort requires multiple teams working together and independently to develop key technologies. Caltech/UCLA Propose to help develop a larger, more automated and flexible test facility, capable of testing and characterizing multiple detector technologies and large arrays. Propose to recruit postdoc and grad student devoted to project.

28. 28 Dark Energy? The study of the nature of dark matter and dark energy, and their effects on the evolution and structure of the universe, are some of the most compelling scientific goals of this century, as recognized by the DOE Office of Science / HEP.

29. 29 Hubble diagram – low z

30. 30 Type Ia Supernovae as Standard Candles Progenitor C/O White Dwarf accreting from companion Just before Chandrasekhar mass, thermonuclear runaway Standard explosion from nuclear physics ("standard candle"). From the luminosity, we can measure the distance from us From the spectrum, we measure how fast they are moving away from us. From this, we construct a Hubble diagram, and infer the expansion rate of the universe, and the history of the expansion rate since the Big Bang (acceleration of the universe).

31. 31 Hubble diagram - SCP 0.2 0.51 In flat universe:  M =0.28 [ .085 stat][ .05 syst] Prob. of fit to  =0 universe: 1% redshift z 0.20.4 0.6 0.8 1.0

32. 32 Supernova Cosmology

33. 33 SNAP: The Third Generation

34. 34 Dark Energy Equation of State

35. 35 Dark Energy Exploration with SNAP Current ground based compared with Binned simulated data and a sample of Dark energy models

36. 36 Science of Weak Lensing Mapping of the distribution of Dark Matter on various scales Measurement of the evolution of structures Measurement of cosmological parameters, breaking degeneracies present in other methods (SNe, CMB) Explore models beyond the standard cosmological model (  CDM) From the statistics of the shear field, weak lensing provides : Jain, Seljak & White 1997, 25’x25’, SCDM

37. 37 Predicted SNAP Constraints on Dark Energy  Redshift information  full 3-dimensional lensing information (“Tomography”) will improve constraints on w by a factor of 2 SNAP wide (300 deg 2 ) Complementarity of weak lensing and SNe !! See e.g. Astro-ph/0507458 SNAP Collaboration, Albert, et al. 2005 (SNAP general) Astro-ph/0507459 SNAP Collaboration, Albert, et al. 2005 (SNe constraints) Astro-ph/0507460 SNAP Collaboration, Albert, et al. 2005 (WL constraints)

38. 38 Test dewar at Caltech OA Mount & enclosure on electronics box (dewar absent for clarity) Molybdenum detector heater plate Flex circuit potted into cap of light tight enclosure with silver filled epoxy:  Thermal ground  Light seal Labyrinth for light seal IR LEDs for flat-field illumination

39. IR Detectors Plenary Session Roger Smith 2006-06-01 Collaboration Meeting

40. 40Persistence Integrated charge (SUR readout) after various length exposures with flux adjusted to give ~same final charge. A matched dark exposure has been subtracted from each ramp to remove reset anomaly as well as dark current. Both amplitude and time constant increase with duration of stimulus. Persistence > dark current for many exposures, so need:  Dynamic bad pixel mapping  Model for correction once below acceptable limits  mitigation through clocking, biasing …if possible. Ubiquitous problem. Manufacturers seem able to make it worse but not better

41. 41 Zero Point Drift Reference pixel subtraction works. We have learned how to do it optimally. Significantly deconstrains bias voltage stability, though we have yet to demonstrate the worst tolerable case. Idea to be tested soon: Since coupling factors differ, correct for thermal drift (pixel by pixel) using temperature sensor on package, then correct only for bias voltages using reference pixels (unity coupling).

42. 42 Pixel operability Example: tails in H2RG-103 noise histograms Need more sophisticated specifications which recognize that most parameters exhibit a distribution of values. A more elaborate operability spec would be a good start. Consider sensitivity, noise, persistence, dark current, etc. How should operability thresholds be set. Need simulations? 4% of pixels > 2*mode: quarter due to hot pixels quarter due to detector noise. half due to mux RTS

43. 43 Cosmic Rays Cosmic ray events with and without CdZnTe substrate removed from H2RG detectors. Each image is the differences of consecutive dark frames so dark current and hot pixels are suppressed and particle events can be positive or negative. Substrate intact, 3000s exposures 2.5  m H2RG-019 @80K Substrate intact, 300s exposures 1.7  m H2RG-038 @100K Substrate removed, 3000s exposures 1.7  m H2RG-019 @100K Raytheon have successfully removed most of substrate but we need all removed. Process refinement continues.

44. 44 Gain and Linearity R. Smith and his team have done thorough studies of readout controller noise, and determined a set of operating parameters to minimize the noise. Caltech SURFs Ben Olsen and Jeff Naecker are measuring detector response to LED light sources, determining the gain, linearity, and stability of the light response as functions of  light flux  exposure time  LED wavelength  detector temperature  detector bias voltages  area on the detector  readout method (Fowler sampling) Tight control of  LED stability  pixels hit with cosmic rays  zero-point drifts  signal chain noise, non-linearity

45. 45 Interpixel Capacitance measured by 2D autocorrelation of shot noise in flat fields measured by 2D autocorrelation of shot noise in flat fields Inter-pixel capacitances are similar for Rockwell and Raytheon, but Rockwell suffers twice the crosstalk, since it has half input capacitance. For Rockwell, if all the light fell on one pixel, 18% of signal would be redistributed to immediate neighbors. (9% for Raytheon). U Michigan